Compiler Optimization of copy construction? - c++

The following code:
#include <iostream>
using namespace std;
struct A {
A(int a) { cout << "cast "; }
A(const A& a) { cout << "copy " ;}
};
int main () {
int x = 0;
A a = x;
return 0;
}
gives the output cast, while I expected the output to be: cast copy.
Is it a compiler optimization, or a simple misunderstanding of what's going on?
Thanks in advance!

An implicit conversion from int to A through the constructor A(int i) is happening over here.
So, I guess, compiler does some optimization.
If you use explicit keyword for first constructor, it won't compile.

Related

Why is object member value changing between 2 getter calls

I am getting unexpected value in second getter call which looks wrong to me, any specific reason for this happening?
#include<iostream>
using namespace std;
class Test {
public:
int &t;
Test (int x):t(x) { }
int getT() { return t; }
};
int main() {
int x = 20;
Test t1(x);
cout << t1.getT() << " ";
cout << t1.getT() << endl;
return 0;
}
The problem here is that your code results in undefined behavior.
The constructor of Test does not take a reference to an int but a copy, and due to int x only being a temporary copy which is not guaranteed to live until your second function call you will end up with undefined behavior.
You would have to change your constructor to the following to make the code work properly:
Test(int& x) : t(x) {}
Now the reference you're working with in Test will be the same x as defined in main

unable to include multiple expressions in initialization of a variable c++

It is stated on the site cppreference.com, something like that
For each declarator, the initializer may be one of the following:
( expression-list ) (1)
= expression (2)
{ initializer-list } (3)
comma-separated list of arbitrary expressions and braced-init-lists in parentheses
But in my code
int main(){
int a,b=5,c(a,b);
return 0;
}
when I try to compile, the following error occurs
...error: expression list treated as compound expression in initializer [-fpermissive]
My question is, if list of multiple expressions is allowed in such style of initialization, then why the compiler is not accepting it with variable c?
What am I missing?
All right, let's look at this:
int main(){
int a,b=5,c(a,b);
return 0;
}
What do you expect c(a,b) to actually do?
Let's simplify this just slightly:
int main(){
int a,b=5;
int c(a,b);
return 0;
}
This will generate the same syntax error, but it now stands alone. So...
Your code would work if there were a constructor for int that took two ints as parameters. This would also compile:
int c(int a, int b);
But in that case, you're actually defining a function.
Also, this works:
int main() {
int a = 5;
int b = 10;
int c(b);
std::cout << "C == " << c << std::endl;
}
That works because an int can be initialized from a single int. But you're getting an error because you can't initialize an int from two other ints.
This works:
#include <iostream>
class MyClass {
public:
MyClass(int a, int b): value(a + b) {}
int value;
};
int main() {
int a = 5;
int b = 10;
MyClass c(a, b);
std::cout << "C == " << c.value << std::endl;
}
And maybe that's what the article you read was trying to tell you. Note: cpppreference is NOT a good site for learning C++. Get a good book.

Why doesn't compiler raise an error when copy assignment function doesn't return anything?

I have run this code on VS2013 and Dev-C++ but when the copy assignment doesn't return anything while actually it should, the compiler doesn't raise any error, please help me to explain this.
#include <iostream>
using namespace std;
class sample
{
public:
sample()
{
cout << "X::X()" << endl;
}
sample(sample const &)
{
cout << "X::X( X const & )" << endl;
}
sample& operator=(sample const &)
{
cout << "X::operator=(X const &)" << endl;
}
};
sample f()
{
sample tmp;
return tmp;
}
int main()
{
int a;
sample x = f();
cin >> a;
return 0;
}
if I change to:
sample x;
x = f();
VS2013 compiler will raise an error like:
Error 1 error C4716: 'sample::operator=' : must return a value c:\users\xxx\desktop\test\test\main.cpp 33 1 Test
Strictly speaking, the compiler isn't required to diagnose this error, since a sufficiently twisted function could make that very difficult or even impossible. However, a decent compiler should be able to give a warning in a simple case like this; for example, GCC will if you specify -Wreturn-type or -Wall.
From what you say, it sounds like Visual Studio, with whatever settings you're using, is only diagnosing this if the function is called. Your first snippet of code performs copy-initialisation (calling the copy constructor), but no assignment, so the assignment operator isn't called.

No call to the copy constructor

Consider this:
#include <iostream>
struct A{
A(){
std::cout << "Create empty A" << std::endl;
}
A(const A& a){
// Why is this never called??
std::cout << "Never called" << std::endl;
}
};
A genA() {
A a;
return a;
}
int main(int argc, const char *argv[])
{
A a(genA()); // Expected to call copy constructor
return 0;
}
Why is the copy constructor not called?
What should I do if I want to ensure that "Never called" is printed on the screen every time I copy A.
This is called as Return value optimization.
Compiler can optimize your code so that it bulds the object directly in the location where it would have been copied too. Thus there will be no reason to use the copy constructor.
Note: The standard explicitly allows it do so.
One way you can force the compiler to avoid RVO is to not return by value - e.g.
#include <iostream>
#include <memory>
struct A{
A() {
std::cout << "Create empty A" << std::endl;
}
A(const A& a) {
// This will be called now
std::cout << "Never called" << std::endl;
}
};
std::auto_ptr<A> genA() {
return std::auto_ptr<A>(new A);
}
int main(int argc, const char *argv[])
{
A a(*(genA().get())); // this will trigger the copy-ctor
return 0;
}
Though this is a horrible hack. I would instead ask why you want to do this? and if you are hoping to put some side-effects into the constructor, urge you to think otherwise.
Copy constructor invocations can be elided (even if they contain side effects), the standard allows that ([12.2]).
EDIT:
I suggest that you do not try to fight it in real world code.
If you just want to see the copy ctor executed in some example/turorial code, then it usually helps not to compile with optimization. Some compilers even have switches that help to avoid that. For GCC it should be -fno-elide-constructors.

Dependent variables in C++?

I tried asking before but I wasn't very clear so I'm re-asking it.
I want to have a variable that depends on the value of another variable, like b in this example:
int main(){
int a;
dependent int b=a+1; //I'm just making this up
a=3;
cout << b; //prints 4
a=4;
cout << b; //prints 5
}
Of course, this does not exist in C++, but this is what I want.
So instead I tried making a function:
int main(){
int a;
int b(){ return a+1; } //error
a=3;
cout << b(); //would print 4 if C++ allowed nested functions
a=4;
cout << b(); //would print 5 if C++ allowed nested functions
}
The above doesn't work because C++ doesn't allow nested functions.
I can only make functions outside of main(), like this:
int b(){
return a+1; //doesn't work because a is not in scope
}
int main(){
int a;
a=3;
cout << b();
a=4;
cout << b();
}
But this does not work because a is not in the same scope as b(), so I would have to pass a as a parameter and I don't want to do that.
Are there any tricks to get something similar to a dependent variable working in C++?
What you need is a closure. If you can use C++ 0x features, you are in luck. Otherwise, you can define one manually:
#include <iostream>
using namespace std;
struct B
{
const int & a;
B(const int & a) : a(a) {}
// variable syntax (Sean Farell's idea)
operator int () const { return a + 1; }
// function syntax
int operator () () const { return a + 1; }
};
int main()
{
int a;
B b(a);
a = 3;
cout << b << '\n'; // variable syntax
a = 4;
cout << b() << '\n'; // function syntax
}
You can also define B inside main, but some compilers would not like it.
The C++ 0x lambda syntax looks like this:
auto b = [&]() { return a + 1; }
The [&] means that the lambda captures local variables by reference.
If you're using C++0x (GCC 4.5+, Visual C++ 2010), you can use lambdas:
int a = 5;
auto b = [&a]{ return a + 1; };
std::cout << b() << std::endl;
Depending on what you're doing, though, there are probably cleaner solutions - possibly some variation of the classic "method that takes in 'a' and returns 'b'"
You could define a class that had a member a, and then a function b() that returned the value of a+1. A basic implementation would be something like:
class Dependent {
public:
Dependent(void) { m_value = 0; }
void set(int value) { m_value = value; }
int b(void) { return(m_value + 1); }
private:
int m_value;
};
int main(){
Dependent a;
a.set(3);
cout << a.b();
a.set(4);
cout << a.b();
}
You could add operator overloading as appropriate to make it work more like normal integers if you so desired.
This is possible if you use lambda functions (c++0x), because they can capture local variables.
Example:
int main()
{
int a;
auto f = [&] () -> int { return a + 1; };
a = 3;
std::cout << f() << std::endl;
a = 4;
std::cout << f() << std::endl;
return 0;
}
Result:
4
5
(See http://ideone.com/MlzX7 for proof)
A simple approach is to use pre-processor macros, nothing C++ specific about it though:
#define b ((a)+1)
int main(){
int a;
a=3;
cout << b;
a=4;
cout << b;
}
#undef b
Are you OK using C++0x ? if yes,
int main()
{
int a = 10;
auto b = [&a]() -> int { return a + 1; };
cout << b() << endl;
}
Since, it is not tagged with c++0x, you can use nested classes instead of nested functions. This column from Herb sutter would help you for existing c++. http://www.gotw.ca/gotw/058.htm
The above doesn't work because C++ doesn't allow nested functions.
You can simulate that using nested structure. In C++0x you can make use of lambda function, which provides the same means of function inside function.
Define a class called LinkedInt or something that behaves like an int, but has a RelatedTo relationship on itself and an additional member that is a function pointer to the function to evaluate when computing the integer's value. Pretty straightforward. Let me know if you need some pointers on the coding.
The short answer is that OOP is more than enough to bury this problem.
I want to have a variable that depends on the value of another
variable, like b in this example:
I see you just need a reference variable:
int a;
int &b =a;
a=10;
cout << b; // 10
Why C++0x lambdas do come for this, I dont understand.